The New Engineer
Sign inStart free
All paths

Agent Systems Engineering

Career route

Engineer tool-using agent loops with explicit context, memory, orchestration, safety, evaluation, and production control.

~14h14 lessons

1. Common Core

5 lessons

Build a typed tool surface and a bounded agent loop.

You'll build: A deterministic agent loop with typed tools, validation, stop conditions, and a transcript.

  1. The Tool Interface — Why Agents Need Structured I/O~1h
  2. Tool Schema Design — Naming, Descriptions, Parameter Constraints~1h
  3. MCP Fundamentals: Stateless Requests and JSON-RPC~1h
  4. The Agent Loop: Observe, Think, Act~1h
  5. Tool Use and Function Calling~1h

2. Role Practice

4 lessons

Make context, memory, state, and orchestration explicit and testable.

You'll build: A stateful workflow with a context budget, memory policy, checkpoints, and recovery paths.

  1. Context Engineering: Windows, Budgets, Memory, and Retrieval~2h
  2. Agent Memory — Virtual Context and Memory Paging~1h
  3. Stateful Graph Orchestration — Durable Execution and Checkpoints~1h
  4. Orchestration Patterns: Supervisor, Swarm, Hierarchical~1h

3. Proof Project

3 lessons

Stress the system against realistic failures and unsafe inputs before release.

You'll build: An observable agent system with adversarial tests, trajectory evaluations, and explicit release gates.

  1. Failure Modes: Why Agents Break~1h
  2. Prompt Injection and the PVE Defense~1h
  3. Eval-Driven Agent Development~1h

4. Interview and Readiness Evidence

2 lessons

Show that the runtime can be operated from controls and trace evidence.

You'll build: A case study with architecture, failure evidence, runtime controls, traces, and justified tradeoffs.

  1. Production Runtimes: Queue, Event, Cron~1h
  2. Agent Observability: Langfuse, Phoenix, Opik~1h